What do you like best?
With bokeh, I can create interactive graphs which gave a whole new dimension to my performance testing reports. Not only they look beautiful, but they let me illustrate concepts in an interactive way, without generating different graphs which are only zoomed in versions or simple numbers: I can show the numbers by hovering over the generated graph instead
What do you dislike?
it took me a while to get the right configuration. Matplotlib works almost out of the box, bokeh takes very little to produce a nice graph, but quite some work to get exactly what you want. Anyway, this is justified by the fact the final outcome is more captivating than Matplotlib and interactive
Recommendations to others considering the product
Watch out for the NumPy dependency: it might be tricky to install on some OS.
Apart from this, installation was smooth and the learning curve shallow enough to start producing logs within the first few minutes of usage
What business problems are you solving with the product? What benefits have you realized?
I managed to automate the full performance testing process by replacing the old tool we were using. I don't need to rely anymore on a black box third party, which produced only certain graphs, I can customize the output I want and format them in a more intelligible way
In addition, inspection for individual results (over millions of samples) is much easier